This paper is concerned with the traveling wave fronts for a lattice dynamical system with global interaction, which arises in a single species in a 2D patchy environment with infinite number of patches connected locally by diffusion and global interaction by delay. We prove that all non-critical traveling wave fronts are globally exponentially stable in time, and the critical traveling wave fronts are globally algebraically stable by the weighted energy method combined with the comparison principle and the discrete Fourier transform.
Citation: Cui-Ping Cheng, Ruo-Fan An. Global stability of traveling wave fronts in a two-dimensional lattice dynamical system with global interaction[J]. Electronic Research Archive, 2021, 29(5): 3535-3550. doi: 10.3934/era.2021051
[1] | Cui-Ping Cheng, Ruo-Fan An . Global stability of traveling wave fronts in a two-dimensional lattice dynamical system with global interaction. Electronic Research Archive, 2021, 29(5): 3535-3550. doi: 10.3934/era.2021051 |
[2] | Xiaoli Wang, Peter Kloeden, Meihua Yang . Asymptotic behaviour of a neural field lattice model with delays. Electronic Research Archive, 2020, 28(2): 1037-1048. doi: 10.3934/era.2020056 |
[3] | Ting Liu, Guo-Bao Zhang . Global stability of traveling waves for a spatially discrete diffusion system with time delay. Electronic Research Archive, 2021, 29(4): 2599-2618. doi: 10.3934/era.2021003 |
[4] | Lianbing She, Nan Liu, Xin Li, Renhai Wang . Three types of weak pullback attractors for lattice pseudo-parabolic equations driven by locally Lipschitz noise. Electronic Research Archive, 2021, 29(5): 3097-3119. doi: 10.3934/era.2021028 |
[5] | Min Li . Long-wavelength limit for the Green–Naghdi equations. Electronic Research Archive, 2022, 30(7): 2700-2718. doi: 10.3934/era.2022138 |
[6] | Denghui Wu, Zhen-Hui Bu . Multidimensional stability of pyramidal traveling fronts in degenerate Fisher-KPP monostable and combustion equations. Electronic Research Archive, 2021, 29(6): 3721-3740. doi: 10.3934/era.2021058 |
[7] | Najmeddine Attia, Ahmed Ghezal . Global stability and co-balancing numbers in a system of rational difference equations. Electronic Research Archive, 2024, 32(3): 2137-2159. doi: 10.3934/era.2024097 |
[8] | Chang Hou, Hu Chen . Stability and pointwise-in-time convergence analysis of a finite difference scheme for a 2D nonlinear multi-term subdiffusion equation. Electronic Research Archive, 2025, 33(3): 1476-1489. doi: 10.3934/era.2025069 |
[9] | Bin Wang . Random periodic sequence of globally mean-square exponentially stable discrete-time stochastic genetic regulatory networks with discrete spatial diffusions. Electronic Research Archive, 2023, 31(6): 3097-3122. doi: 10.3934/era.2023157 |
[10] | Amira Khelifa, Yacine Halim . Global behavior of P-dimensional difference equations system. Electronic Research Archive, 2021, 29(5): 3121-3139. doi: 10.3934/era.2021029 |
This paper is concerned with the traveling wave fronts for a lattice dynamical system with global interaction, which arises in a single species in a 2D patchy environment with infinite number of patches connected locally by diffusion and global interaction by delay. We prove that all non-critical traveling wave fronts are globally exponentially stable in time, and the critical traveling wave fronts are globally algebraically stable by the weighted energy method combined with the comparison principle and the discrete Fourier transform.
Lattice differential equations (LDEs) are systems of ordinary differential equations with a discrete spatial structure, which can naturally arise in various fields, such as image processing, neural networks, patten recognition and chemical reaction theory. These can be seen in [9,14,29,37] and the references therein. Recently, there is a particular interest on studying the species population living in a patchy environment consisting of all integer nodes, see [7,8,34,35].
Inspired by Bates [1], Chow [9], Weng et al.[34] and many other excellent survey papers, the authors in [7] considered a single-species population with two-age classes distributed over a patchy environment consisting of all integer nodes of a 2D lattice and derived the following system:
dwk,j(t)dt=Dm[wk+1,j(t)+wk−1,j(t)+wk,j+1(t)+wk,j−1(t)−4wk,j(t)]−dmwk,j(t)+μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b(wk−l,j−q(t−r)), | (1.1) |
where
βα(l)=2e−2α∫π0cos(lω1)e2αcosω1dω1, |
γα(l)=2e−2α∫π0cos(lω2)e2αcosω2dω2, |
for any
Assume the birth function
The authors [7] studied the existence of the asymptotic speed of propagation, the existence of monotone traveling waves and the minimal wave and its relation with the asymptotic speed of propagation. Recently, Xu [35] further showed that for any given admissible speed, all the wave profiles propagating toward a fixed direction of (1.1) have the same asymptotic behavior when they approach the limiting states, which plays a very important role in the stability of traveling waves. Meanwhile, in the past decades, there are various surveys focusing on the existence, uniqueness, asymptotic behavior and stability of traveling waves for LDEs and its continuum RDEs([1,2,3,4,6,7,8,9,11,13,16,17,19,20,21,22,37,38,39]).
In this paper, we are concerned with the stability of traveling waves of (1.1) under the assumptions
wk,j(t)|t=s=w0k,j(s),for s∈[−r,0],k,j∈Z, | (1.2) |
where
w0k,j(s)→0,for all s∈[−r,0],as kcosθ+jsinθ→−∞;w0k,j(s)→w+,for all s∈[−r,0],as kcosθ+jsinθ→∞, | (1.3) |
we prove that the global solution
As we know, many techniques are developed to investigate the stability of traveling waves such as the spectral analysis method, the weighted energy method ([5,8,12,14,15,18,23,24,25,26,27,28,30,37,38]) and the comparison principle combining the squeezing technique ([4,6,19,20,21,22,31]). Recently, Mei et al. [25] and Huang et al.[15] obtained the global stability of traveling wave fronts with noncritical speed and critical speed of nonlocal reaction-diffusion equations via the weighted energy method together with the comparison principle and Green function or Fourier transform. Zhang [38] applied this method to nonlocal LDEs in one dimension. However, there seems to be not much progress on the stability of traveling waves of system (1.1).
Particularly in [8], we only consider the case that the immature population is non-mobile, that is
dwk,j(t)dt=Dm[wk+1,j(t)+wk−1,j(t)+wk,j+1(t)+wk,j−1(t)−4wk,j(t)]−dmwk,j(t)+μb(wk,j(t−r)). | (1.4) |
In monostable case, using weighted energy method, we derived that the Cauchy problem of (1.4) converges to a traveling wavefront when the initial perturbation around the wave is suitably small in a weighted norm. Due to the limitation of the key inequality, the result only holds for
The outline of this paper is as follows: in Section 2, we introduce some preliminaries, recall the result on the existence of traveling wave fronts of (1.1) and present our main result. Section 3 is devoted to the global stability of traveling wave fronts by using the weighted energy method combined with the semi-discrete Fourier transform.
Notations. Let
l∞={c={ck,j}k,j∈Z,ck,j∈R;‖c‖l∞:=supk,j∈Z|ck,j|<∞}, |
let
l1={c={ck,j}i∈Z,ck,j∈R;‖c‖l1:=∑k,j∈Z|ck,j|<∞}, |
and denote by
l2={c={ck,j}k,j∈Z,ck,j∈R;‖c‖l2:=(∑k,j∈Z|ck,j|2)12<∞}. |
Further,
‖c‖lpw:=(∑k,j∈Zw(kcosθ+jsinθ)|ck,j|p)1p. |
For any
F[v](ω)=ˆv(ω)=12π∑k∈Z∑j∈Ze−i(kω1+jω2)vk,jω=(ω1,ω2)∈[−π,π]2, |
and the inverse Fourier transform is given by
F−1[ˆv]=12π∫π−π∫π−πei(kω1+jω2)ˆv(ω)dω1dω2k,j∈Z, |
where
A traveling wave of system (1.1) is
cdϕ(x)dx=Dm(ϕ(x+cosθ)+ϕ(x−cosθ)+ϕ(x+sinθ)+ϕ(x−sinθ)−4ϕ(x))−dmϕ(x)+μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b(ϕ(x−lcosθ−qsinθ−cr)), | (2.1) |
where
limx→−∞ϕ(x)=0,limx→∞ϕ(x)=w+. | (2.2) |
Denoting the characteristic equation at the trivial equilibrium
Δ(λ,c;θ)=Dm(eλcosθ+e−λcosθ+eλsinθ+e−λsinθ−4)−cλ−dm+μb′(0)(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)e−(λlcosθ+λqsinθ+λcr). |
It is easy to see that
Lemma 2.1. [7,Lemma 4.2] There exist a pair of
The existence of traveling wave fronts for (2.1) with the boundary condition (2.2) can be easily verified by using the monotone iteration technique combined with the sub-sup solutions, see [7].
Lemma 2.2. [7,Theorem 5.4] Assume
Recently, Xu [35] derived the asymptotic behavior of traveling waves of (1.1), which is the key premise.
Lemma 2.3. [35,Theorem 2.1] Assume
limx→−∞ϕ(x)=(m−x)leλ1(c)x |
for some
Now, define a weight function
ν(x)=e−λ∗(x−x∗), | (2.3) |
where
We now state our main stability result in this paper.
Theorem 2.4. Assume that
dds(w0k,j(s)−ϕ(kcosθ+jsinθ+cs))∈L1([−r,0],l1ν(Z2)), |
then the unique solution
0≤wk,j(t)≤w+,k,j∈Z,t≥0, |
and
wk,j(t)−ϕ(kcosθ+jsinθ+ct)∈C0([0,+∞),l1ν). |
Furthermore, for
supk,j∈Z|wk,j(t)−ϕ(kcosθ+jsinθ+ct)|≤C1(1+t)−2κe−μ0t,t≥0, |
for some constant
supk,j∈Z|wk,j(t)−ϕ(kcosθ+jsinθ+c∗t)|≤C3(1+t)−2κ,t≥0. |
We first consider the following initial value problem:
{dwk,j(t)dt=Dm[wk+1,j(t)+wk−1,j(t)+wk,j+1(t)+wk,j−1(t)−4wk,j(t)]−dmwk,j(t)+μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b(wk−l,j−q(t−r)),wk,j(s)=w0k,j(s)∈C0([−r,0],[0,w+]), | (3.1) |
where
Definition 3.1. A sequence of continuous differentiable functions
dwk,j(t)dt≥(≤)Dm[wk+1,j(t)+wk−1,j(t)+wk,j+1(t)+wk,j−1(t)−4wk,j(t)]−dmwk,j(t)+μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b(wk−l,j−q(t−r)). | (3.2) |
Theorem 3.2. For any given function
W0={w0k,j}k,j∈Z,w0k,j∈C0([−r,0],[0,w+]),k,j∈Z |
and
Note that (3.1) is equivalent to
{wk,j(t)=e−δtwk,j(0)+∫t0e−δs(Gw(s)+Hw(s))ds,wk,j(t)=w0k,j(s)∈C0([−r,0],[0,1]), |
where
Gw(s)=Dm[wk+1,j(s)+wk−1,j(s)+wk,j+1(s)+wk,j−1(s)], |
and
Hw(s)=μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b(wk−l,j−q(s−r)). |
Theorem 3.2 can be verified using an argument in [7,Theorem 3.1] and [33,Lemma 3.2] and we omit it here.
Let
w+k,j(s)=max{w0k,j(s),ϕ(kcosθ+jsinθ+cs)},w−k,j(s)=min{w0k,j(s),ϕ(kcosθ+jsinθ+cs)}, |
for
0≤w−k,j(s)≤w0k,j(s)≤w+k,j(s)≤w+,s∈[−r,0], |
and
0≤w−k,j(s)≤ϕ(kcosθ+jsinθ+cs)≤w+k,j(s)≤w+,s∈[−r,0]. | (3.3) |
Denote
{dw±k,j(t)dt=Dm[w±k+1,j(t)+w±k−1,j(t)+w±k,j+1(t)+w±k,j−1(t)−4w±k,j(t)]−dmw±k,j(t)+μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b(w±k−l,j−q(t−r)),t≥0,w±k,j(s)=w±k,j(s),s∈[−r,0]. |
By the comparison principle, we have
0≤w−k,j(t)≤wk,j(t)≤w+k,j(t)≤w+,t≥0,0≤w−k,j(t)≤ϕ(kcosθ+jsinθ+ct)≤w+k,j(t)≤w+,t≥0. | (3.4) |
Thus
w−k,j(t)−ϕ(x)≤wk,j(t)−ϕ(x)≤w+k,j(t)−ϕ(x), | (3.5) |
where
Step 1. The convergence of
{zk,j(t)=w+k,j(t)−ϕ(kcosθ+jsinθ+ct),t≥0z0k,j(s)=w+k,j(s)−ϕ(kcosθ+jsinθ+cs),s∈[−r,0], | (3.6) |
According to (3.3) and (3.4), we have
{dzk,j(t)dt−Dm[zk+1,j(t)+zk−1,j(t)+zk,j+1(t)+zk,j−1(t)−4zk,j(t)]+dmzk,j(t)−b′(0)μ(2π)2∞∑l,q=−∞βα(l)γα(q)zk−l,j−q(t−r)=μ(2π)2∞∑l,q=−∞βα(l)γα(q)[b(zk−l,j−q(t−r)+ϕ)−b(ϕ)−b′(ϕ)zk−l,j−q(t−r)]+μ(2π)2∞∑l,q=−∞βα(l)γα(q)[b′(ϕ)−b′(0)]zk−l,j−q(t−r)=μ(2π)2∞∑l,q=−∞βα(l)γα(q)mk,j(t−r)+μ(2π)2∞∑l,q=−∞βα(l)γα(q)nk,j(t−r),zk,j(s)=z0k,j(s), | (3.7) |
where
mk,j(t−r)=b(z+ϕ)−b(ϕ)−b′(ϕ)zk,j(t−r),nk,j(t−r)=[b′(ϕ)−b′(0)]zk,j(t−r), |
with
b′(ϕ)−b′(0)≤0 |
and
mk,j(t−r)=b(z+ϕ)−b(ϕ)−b′(ϕ)zk,j(t−r)=b″(˜ϕ)2!z2≤0, |
where
dzk,j(t)dt−Dm[zk+1,j(t)+zk−1,j(t)+zk,j+1(t)+zk,j−1(t)−4zk,j(t)]+dmzk,j(t)−b′(0)μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)zk−l,j−q(t−r)≤0. |
Let
{dzk,j(t)dt−Dm[zk+1,j(t)+zk−1,j(t)+zk,j+1(t)+zk,j−1(t)−4zk,j(t)]+dmzk,j(t)−b′(0)μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)zk−l,j−q(t−r)=0,ˉzk,j(s)=z0k,j(s),s∈[−r,0]. | (3.8) |
Then according to the comparison principle, we have
0≤zk,j(t)≤ˉzk,j(t),k,j∈Z,t>0. | (3.9) |
Let
{dz∗k,j(t)dt−Dm[eλ∗cosθz∗k+1,j(t)+e−λ∗cosθzk−1,j(t)+eλ∗sinθz∗k,j+1(t)+e−λ∗sinθz∗k,j−1(t)]+(cλ∗+4Dm+dm)z∗k,j(t)=μb′(0)(2π)2∞∑l,q=−∞βα(l)γα(q)e−λ∗(kcosθ+lsinθ+cr)z∗k−l,j−q(t−r),z∗k,j(s)=e−λ∗(kcosθ+jsinθ+cs−x∗)z0k,j(s),s∈[−r,0]. | (3.10) |
By Fourier transformation, one has
F[eλ∗cosθz∗k+1,j(t)]=12π∑k∈Z∑j∈Ze−i(kω1+jω2)eλ∗cosθz∗k+1,j(t)=eλ∗cosθeiω112π∑k∈Z∑j∈Ze−i((k+1)ω1+jω2)z∗k+1,j(t)=eλ∗cosθeiω1ˆz∗(t,ω), |
F[b′(0)μ(2π)2∞∑l,q=−∞βα(l)γα(q)e−λ∗(lcosθ+qsinθ+cr)z∗k−l,j−q(t−r)]=12π∑k∈Z∑j∈Ze−i(kω1+jω2)[b′(0)μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)e−λ∗(lcosθ+qsinθ+cr)z∗k−l,j−q(t−r)]=b′(0){μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)e−λ∗(lcosθ+qsinθ+cr)e−i(lω1+qω2)}ˆz∗(t−r,ω), |
where
{dˆz∗(t,ω)dt+A(ω)ˆz∗(t,ω)=B(ω)ˆz∗(t−r,ω),ˆz∗(s,ω)=ˆz∗0(s,ω),s∈[−r,0], | (3.11) |
where
A(ω)=cλ∗+4Dm+dm−Dm[eλ∗cosθeiω1+e−λ∗cosθe−iω1+eλ∗sinθeiω2+e−λ∗sinθe−iω2], |
B(ω)=b′(0){μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)e−λ∗(lcosθ+qsinθ+cr)e−i(lω1+qω2)}, |
ˆz∗0(s,ω)=12π∑k∈Z∑j∈Ze−i(kω1+jω2)e−λ∗(kcosθ+jsinθ+cs−x∗)z0k,j(s). |
In order to obtain the decay estimates of
Lemma 3.3. Let
{dx(t)dt+k1x(t)=k2x(t−r),t>0,r>0,x(s)=x0(s),s∈[−r,0]. |
Then
x(t)=e−k1(t+r)ek3trx0(−r)+∫0−re−k1(t−s)ek3(t−r−s)r[z′0(s)+k1z0(s)]ds, |
where
ek3tr={0,−∞<t<−r,1,−r≤t<0,1+k3t1!,0≤t<r,1+k3t1!+k23(t−r)22!,r≤t<2r,⋮⋮1+k3t1!+k23(t−r)22!+⋯+km3[t−(m−1)rm!]m,(m−1)r≤t<mr,⋮⋮ |
and
{ddtx(t)=k3x(t−r),t≥0,x(s)≡1,s∈[−r,0]. |
Furthermore, it is pointed in [36] that when
e−k1tek3tr≤Ce−ϵ1(k1−k2)t,t>0. | (3.12) |
In view of Lemma 3.3, the solution of (3.11) can be given as follows:
ˆz∗(t,ω)=e−A(ω)(t+r)eB(ω)trˆz∗0(−r,ω)+∫0−re−A(ω)(t−s)eB(ω)(t−r−s)r[∂sˆz∗0(s,ω)+A(ω)ˆz∗0(s,ω)]ds, | (3.13) |
where
z∗k,j(t)=12π∫π−π∫π−πei(kω1+jω2)e−A(ω)(t+r)eB(ω)trˆz∗0(−r,ω)dω1dω2+12π∫π−π∫π−π∫0−rei(kω1+jω2)e−A(ω)(t−s)eB(ω)(t−r−s)r[∂sˆz∗0(s,ω)+A(ω)ˆz∗0(s,ω)]dsdω1dω2, | (3.14) |
Let
Pk,j(t)=12π∫π−π∫π−πei(kω1+jω2)e−A(ω)(t+r)eB(ω)trˆz∗0(−r,ω)dω1dω2, |
and
Qk,j(t)=12π∫π−π∫π−π∫0−rei(kω1+jω2)e−A(ω)(t−s)eB(ω)(t−r−s)r[∂sˆz∗0(s,ω)+A(ω)ˆz∗0(s,ω)]dsdω1dω2. |
we first give an estimate of
‖Pk,j(t)‖≤12π∫π−π∫π−π‖e−A(ω)(t+r)‖‖eB(ω)tr‖|ˆz∗0(−r,ω)|dω1dω2. | (3.15) |
For the estimation of
‖e−A(ω)t‖=e−c1t|exp{Dmt(eλ∗cosθeiω1+e−λ∗cosθe−iω1+eλ∗sinθeiω2+e−λ∗sinθe−iω2)}|=e−c1texp{Dmt[cos(ω1)(eλ∗cosθ+e−λ∗cosθ)+cos(ω2)(eλ∗sinθ+e−λ∗sinθ)]}=e−k0texp{−Dmt[(eλ∗cosθ+e−λ∗cosθ)(1−cosω1)+(eλ∗sinθ+e−λ∗sinθ)(1−cosω2)]}=e−k1t, |
where
c1=cλ∗+4Dm+dm, |
k0=c1−Dm[eλ∗cosθ+e−λ∗cosθ+eλ∗sinθ+e−λ∗sinθ], |
and
k1=k0+Dm[(eλ∗cosθ+e−λ∗cosθ)(1−cosω1)+(eλ∗sinθ+e−λ∗sinθ)(1−cosω2)] |
Meanwhile, due to
B(ω)=b′(0){μ(2π)2∞∑l,q=−∞βα(l)γα(q)e−λ∗(lcosθ+qsinθ+cr)e−i(lω1+qω2)} |
|B(ω)|≤μb′(0)(2π)2∞∑l,q=−∞βα(l)γα(q)e−λ∗(lcosθ+qsinθ+cr):=k2 |
‖B(ω)‖≤|BeA(ω)r|≤k2ek1r:=k3, |
we have
|eB(ω)tr|≤ek3tr. |
Then
|Pk,j(t)|≤12π∫π−π∫π−πe−k1(t+r)ek3tr|ˆz∗0(−r,ω)|dω1dω2. | (3.16) |
Since when
k1=k0+Dm[(eλ∗cosθ+e−λ∗cosθ)(1−cosω1)+(eλ∗sinθ+e−λ∗sinθ)(1−cosω2)]=cλ∗+dm−Dm[eλ∗cosθ+e−λ∗cosθ+eλ∗sinθ+e−λ∗sinθ−4]+Dm[(eλ∗cosθ+e−λ∗cosθ)(1−cosω1)+(eλ∗sinθ+e−λ∗sinθ)(1−cosω2)]≥μb′(0)(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)e−λ∗(lcosθ+qsinθ+cr)=k2, |
(3.12) and (3.16) imply that
|Pk,j(t)|≤C2π∫π−π∫π−πe−ϵ1(k1−k2)t|ˆz∗0(−r,ω)|dω1dω2, | (3.17) |
for a constant
Since
exp{−Dmt[(eλ∗cosθ+e−λ∗cosθ)(1−cosω1)+(eλ∗sinθ+e−λ∗sinθ)(1−cosω2)]}≤exp{−2Dmt(1−cosω1)}exp{−2Dmt(1−cosω2)}=exp{−2Dmt[1−(cosω1+isinω1)]}exp{−2Dmt[1−(cosω2+isinω2)]}=exp{Dmt[eiω1+e−iω1+eiω2+e−iω2−4]}. |
According to Taylor series expansion
eiξ=1+iξ−ξ22!−iξ33!+ξ44!−⋯, |
eiξ+e−iξ=2(1−ξ22!+ξ44!−⋯). |
It follows that
e−iξ+eiξ2≤1−M|ξ|κ+|ξ|κh(ξ), |
where
e−iξ+eiξ2≤1−M1|ξ|κ, for |ξ|≤a0, |
and
e−iξ+eiξ2=cosξ≤1−δ, for |ξ|>a0. |
Let
∫π−π∫π−πexp{−ϵ1Dm[4−(e−iω1+eiω1+e−iω2+eiω2)]t}|ˆz∗0(−r,ω)|dω1dω2≤‖ˆz∗0(−r,ω)‖L∞([−π,π]2)×∫π−πexp{−ϵ1Dm[2−(e−iω2+eiω2)]}dω2∫π−πexp{−ϵ1Dm[2−(e−iω1+eiω1)]}dω1=‖ˆz∗0(−r,ω)‖L∞([−π,π]2)(∫π−πexp{−ϵ1Dm[2−(e−iξ+eiξ)]}dξ)2≤‖ˆz∗0(−r,ω)‖L∞([−π,π]2)(∫|ξ|<a0e−2ϵ1DmM1|ξ|κtdξ+∫ξ∈Ea0e−2ϵ1Dmδtdξ)2≤‖ˆz∗0(−r,ω)‖L∞([−π,π]2)(∫|ξ|<a0e−2ϵ1DmM1|ξ|κtdξ+e−2ϵ1Dmδtm(Ea0))2, | (3.18) |
where
By changing variables
∫|ξ|<a0e−ϵ1DmM1|ξ|κtdξ+e−ϵ1Dmδtm(Ea0)≤t−1κ∫|σ|≤a0t1κe−ϵ1DmM1|σ|κdσ+e−ϵ1Dmδtm(Ea0)≤Ct−1κ, |
for some constant
Then
‖Pk,j(t)‖≤C2πe−ϵ1(k0−k2)t×∫π−π∫π−πexp{−ϵ1Dm[4−(e−iω1+eiω1+e−iω2+eiω2)]t}|ˆz∗0(−r,ω)|dω1dω2≤C‖ˆz∗0(−r,ω)‖L∞([−π,π]2)t−2κe−ϵ1(k0−k2)t≤C‖z∗k,j(−r)‖l1(Z2)t−2κe−ϵ1(k0−k2)t. |
Since
‖Pk,j(t)‖≤C‖z∗k,j(−r)‖l1(Z)(1+t)−2κe−ϵ1(k0−k2)t. |
The following inequality can be obtained similarly
‖Qk,j(t)‖≤12π∫0−r∫π−π∫π−π|e−A(ω)(t−s)||eB(ω)(t−r−s)r||∂sˆz∗0(s,ω)+A(ω)ˆz∗0(s,ω)|dω1dω2ds≤C∫0−r(‖z∗k,j(s)‖l1(Z)+‖ddsz∗k,j(s)‖l1(Z))ds(1+t)−2κe−ϵ1(k0−k2)t. |
Consequently, we get the following decay estimate
‖z∗k,j(t)‖l∞(Z)≤C(1+t)−2κe−ϵ1(k0−k2)t. |
When
‖z∗k,j(t)‖l∞(Z)≤C(1+t)−2κe−ϵ1μ1t, |
where
When
‖z∗k,j(t)‖l∞(Z)≤C(1+t)−2κ, |
due to
According to (3.9),
zk,j(t)≤ˉzk,j(t)=eλ∗(kcosθ+jsinθ+ct−x∗)z∗k,j(t), |
let
Lemma 3.4. For any
‖zk,j(t)‖l∞(H)≤C(1+t)−2κe−ϵ1μ1t |
for some
‖zk,j(t)‖l∞(H)≤C(1+t)−2κ. |
Next we will prove
Lemma 3.5. For
‖zk,j(t)‖l∞(Z2∖H)≤C(1+t)−2κe−μ2t, |
where
0<μ2<min{dm−μb′(w+),ϵ1μ1}; |
‖zk,j(t)‖l∞(Z2∖H)≤C(1+t)−2κ. |
Proof. Let
dzk,j(t)dt−Dm[zk+1,j(t)+zk−1,j(t)+zk,j+1(t)+zk,j−1(t)−4zk,j(t)]+dmzk,j(t)=μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)(b(ϕ+z)−b(ϕ)). |
Since
b(ϕ+z)−b(ϕ)=b′(ϕ)z+b″(˜ϕ)z2≤b′(ϕ)z, |
where
{dzk,j(t)dt−Dm[zk+1,j(t)+zk−1,j(t)+zk,j+1(t)+zk,j−1(t)−4zk,j(t)]+dmzk,j(t)≤μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b′(ϕ)zk−l,j−q(t−r),zk,j(t)|kcosθ+jsinθ<x∗−ct≤C2(1+t)−2κe−ϵ1μ1t,zk,j(t)|t=s=z0k,j(s),s∈[−r,0],k,j∈Z, |
for some positive constant
˜z(t)=C3(1+r+t)−2κe−μ2t, |
for
dm−μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b′(ϕ(x−lcosθ−qsinθ−cr))≥ϵ0[dm−μb′(w+)], | (3.19) |
where
d˜z(t)dt+dm˜z(t)−μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b′(ϕ)˜z(t−r)=C3(1+r+t)−2κe−μ2t{dm−μ2−2κ11+r+t−μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b′(ϕ)−[eμ2r(1+t1+r+t)−2κ−1]μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b′(ϕ)}≥C3(1+r+t)−2κe−μ2t{ϵ0[dm−μb′(w+)]−μ2−2κ1(1+r+t)−[eμ2r(1+t1+r+t)−2κ−1]μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b′(ϕ)}≥0, |
by choosing
0<μ2<min{dm−μb′(w+),ϵ1μ1} for c>c∗,μ2=0, for c=c∗. |
Hence, we have
{d˜z(t)dt+dm˜z(t)≥μ(2π)2∞∑l=−∞∞∑q=−∞βα(l)γα(q)b′(ϕ)˜z(t−r),t>l0r,x>x∗,˜z(t)|x=x∗=C3(1+r+t)−2κe−μ2t≥C2(1+t)−2κe−ϵ1μ1t,˜z(t)=C3(1+r+t)−2κe−μ2t>z0k,j(t),t∈[−r,l0r],k,j∈Z. |
Therefore, by the comparison principle, we can get
zk,j(t)≤˜z(t). |
The proof is complete.
It then follows from Lemmas 3.4 and 3.5 that
Lemma 3.6. For any
‖w+k,j(t)−ϕ(x)‖l∞(Z2)≤C(1+t)−2κe−μ2t, |
where
0<μ2<min{dm−μb′(w+),ϵ1μ1}; |
‖w+k,j(t)−ϕ(x)‖l∞(Z2)≤C(1+t)−2κ. |
Step 2. The convergence of
By a similar argument as in Step 1, the proof can be done.
Lemma 3.7. For any
‖ϕ(x)−w−k,j(t)‖l∞(Z2)≤C(1+t)−2κe−μ2t, |
where
0<μ2<min{dm−μb′(w+),ϵ1μ1}; |
‖ϕ(x)−w−k,j(t)‖l∞(Z2)≤C(1+t)−2κ. |
Step 3: The convergence of
Lemma 3.8. For any
‖wk,j(t)−ϕ(x)‖l∞(Z2)≤C(1+t)−2κe−ϵ1μt, |
where
‖wk,j(t)−ϕ(x)‖l∞(Z2)≤C(1+t)−2κ. |
The authors thank the anonymous referee for their valuable comments and suggestions that help the improvement of the manuscript. And the first author was supported by the NSFs of China (No.11301121, No. 11772203).